1
00:00:00,320 --> 00:00:03,480
Speaker 1: Together do for c Allen, an entire class of Lincoln

2
00:00:03,600 --> 00:00:06,840
Speaker 1: University students are being retested because it has suspected that

3
00:00:06,880 --> 00:00:09,280
Speaker 1: some of them use AI to cheat. More than one

4
00:00:09,360 --> 00:00:12,280
Speaker 1: hundred students are now going to have to orally explain

5
00:00:12,360 --> 00:00:14,600
Speaker 1: the code that they wrote for their exam to the

6
00:00:14,600 --> 00:00:17,840
Speaker 1: professor to ensure that they understand their work. Chad Hewitt

7
00:00:17,840 --> 00:00:21,439
Speaker 1: is the provist professor at Lincoln University in christ Church

8
00:00:21,480 --> 00:00:25,200
Speaker 1: and with us now high Chad, Hello, out of the

9
00:00:25,239 --> 00:00:28,400
Speaker 1: one hundred, how many students are suspected of using AI?

10
00:00:29,720 --> 00:00:32,960
Speaker 2: I'm not sure about that. The academic canvolve do it.

11
00:00:33,080 --> 00:00:37,120
Speaker 2: The course has just identified a high number of suspected cases.

12
00:00:38,040 --> 00:00:40,960
Speaker 2: As a consequence, the easiest thing to do is to

13
00:00:41,000 --> 00:00:44,800
Speaker 2: bring the students in and to ask them questions.

14
00:00:45,960 --> 00:00:48,279
Speaker 1: Is it fair to make the whole class reset it

15
00:00:48,520 --> 00:00:50,320
Speaker 1: rather than just the ones you guys suspect?

16
00:00:51,400 --> 00:00:54,040
Speaker 2: Well, I think the challenge is to get a normative

17
00:00:54,120 --> 00:00:59,760
Speaker 2: understanding of the understanding of the assignment, the mechanisms by

18
00:00:59,760 --> 00:01:03,400
Speaker 2: which the students wrote the code. I think it's not

19
00:01:03,680 --> 00:01:07,039
Speaker 2: an onerous task to ask the students questions.

20
00:01:07,160 --> 00:01:10,080
Speaker 1: Okay, should this not have been how you tested them

21
00:01:10,120 --> 00:01:12,319
Speaker 1: in the first place? With oral exams.

22
00:01:13,560 --> 00:01:18,520
Speaker 2: No, in this instance, for fintech, they're very interested in

23
00:01:18,560 --> 00:01:23,039
Speaker 2: the student's ability to write code. We won't know how

24
00:01:23,040 --> 00:01:26,959
Speaker 2: the students undertake the work, so usually you would do that.

25
00:01:27,200 --> 00:01:33,160
Speaker 2: The students were very clearly identified that the desire was

26
00:01:33,200 --> 00:01:36,279
Speaker 2: to understand their work and how they wrote the code,

27
00:01:36,280 --> 00:01:39,399
Speaker 2: with the expectation that some students would make mistakes and

28
00:01:39,440 --> 00:01:43,479
Speaker 2: therefore you learn through those mistakes. In this case, there

29
00:01:43,480 --> 00:01:47,840
Speaker 2: were a very large number of students that indicated a

30
00:01:47,920 --> 00:01:51,720
Speaker 2: high level of proficiency, which is unexpected.

31
00:01:52,840 --> 00:01:55,320
Speaker 1: So okay, let me just try to understand what you

32
00:01:55,360 --> 00:01:57,560
Speaker 1: just said. So if it was the case that only

33
00:01:57,600 --> 00:02:00,240
Speaker 1: a few students had cheated, you'd let its life. But

34
00:02:00,280 --> 00:02:02,000
Speaker 1: because lots of them cheated, you're not going to let

35
00:02:02,040 --> 00:02:02,560
Speaker 1: it slide.

36
00:02:03,840 --> 00:02:07,680
Speaker 2: No, we would not let it slide. The challenge, of course,

37
00:02:07,800 --> 00:02:11,040
Speaker 2: is finding out and being able to detect the use

38
00:02:11,080 --> 00:02:12,960
Speaker 2: of generative artificial intelligence.

39
00:02:13,040 --> 00:02:15,040
Speaker 1: Yeah, I know, but Chad, come on, I mean, the

40
00:02:15,080 --> 00:02:17,720
Speaker 1: thing is is, if kids are using computers, every single

41
00:02:17,760 --> 00:02:21,080
Speaker 1: one of them has the option at least of cheating.

42
00:02:21,400 --> 00:02:24,760
Speaker 1: So should we not, as universities, be moving to using

43
00:02:25,040 --> 00:02:26,839
Speaker 1: oral exams more. I mean, sure, get them to write

44
00:02:26,840 --> 00:02:29,120
Speaker 1: the code, but then make the oral exam part of

45
00:02:29,160 --> 00:02:30,880
Speaker 1: it in order to get around the AI.

46
00:02:31,880 --> 00:02:34,600
Speaker 2: Well, that may be a mechanism that we employ in

47
00:02:34,600 --> 00:02:36,840
Speaker 2: the future, but of course this is a course of

48
00:02:36,960 --> 00:02:40,080
Speaker 2: one hundred and twenty students, and you do try and

49
00:02:40,120 --> 00:02:44,920
Speaker 2: identify things that will allow you to measure the performance

50
00:02:44,919 --> 00:02:50,480
Speaker 2: of the course through simple terms. Obviously, in this case,

51
00:02:51,120 --> 00:02:54,760
Speaker 2: resetting all of the students is an unexpected change of events,

52
00:02:54,800 --> 00:02:58,079
Speaker 2: but it's better than actually turning them into the productor

53
00:02:58,160 --> 00:02:59,680
Speaker 2: and having an investigation occur.

54
00:03:00,320 --> 00:03:02,560
Speaker 1: So the problem is basically that when you have this

55
00:03:02,680 --> 00:03:05,040
Speaker 1: many students, it's just too hard to go through the

56
00:03:05,040 --> 00:03:07,560
Speaker 1: whole kerfuffle of oral exams. Yeah.

57
00:03:07,720 --> 00:03:11,520
Speaker 2: No, the oral exams will give you a lesson in

58
00:03:11,600 --> 00:03:14,119
Speaker 2: terms of some material, but it's not actually what we're

59
00:03:14,120 --> 00:03:16,480
Speaker 2: trying to test. We want to know how the students

60
00:03:16,560 --> 00:03:20,680
Speaker 2: think and code. So having them actually do and perform

61
00:03:20,760 --> 00:03:23,600
Speaker 2: the task is critical in terms of no.

62
00:03:24,120 --> 00:03:26,080
Speaker 1: Chaed I think you misunderstand me. I'm not saying that

63
00:03:26,120 --> 00:03:28,360
Speaker 1: you replace one with the other. I just think in

64
00:03:28,400 --> 00:03:31,560
Speaker 1: the age of AI, we're going to have do, aren't we.

65
00:03:31,639 --> 00:03:34,800
Speaker 1: With everything moved towards getting the kids right in front

66
00:03:34,800 --> 00:03:37,360
Speaker 1: of you, explaining that they understand it whatever the subject.

67
00:03:39,040 --> 00:03:44,720
Speaker 2: Possibly. I think there are instances where the assessments can

68
00:03:44,800 --> 00:03:47,800
Speaker 2: be focused where we allow AI use and then we

69
00:03:47,840 --> 00:03:51,760
Speaker 2: expect the students to criticize it. There are other instances

70
00:03:51,800 --> 00:03:54,320
Speaker 2: where we allow the AI use but for it to

71
00:03:54,360 --> 00:03:58,000
Speaker 2: be clearly articulated and declared. In this case, the students

72
00:03:58,040 --> 00:03:59,960
Speaker 2: were expected to use their own brand.

73
00:04:00,720 --> 00:04:03,880
Speaker 1: Okay, Chad, thanks very much, appreciate it, Chad Hewitt, who's

74
00:04:03,920 --> 00:04:07,040
Speaker 1: the provost professor at Lincoln University.

75
00:04:07,840 --> 00:04:11,000
Speaker 2: For more from Heather Duplessy Allen Drive, listen live to

76
00:04:11,120 --> 00:04:14,160
Speaker 2: news talks. It'd be from four pm weekdays, or follow

77
00:04:14,200 --> 00:04:15,960
Speaker 2: the podcast on iHeartRadio